Teledentistry is becoming urgent in the community and also has many favorable conditions for implementation thanks to the strong development of artificial intelligence today. In this paper, we present a large dataset labeled for common dental diseases. Specifically, we focus on two types of diseases including tartar and tooth decay. We build a dataset of 9613 images (after applying data augmentation) with three basic labels including “calculus”, “caries” and “healthy”. We also train the YOLO v11 model on the proposed dataset and develop it into a website-based application. The results show that the YOLO v11 model achieves impressive performance on the proposed dataset for the application, receiving positive feedback from users for accurate predictions and integrating LLM models for more creative and user-friendly predictions.